THE RELATIONSHIP BETWEEN ORGANIZATIONAL CULTURE

RESEARCH NOTE THE RELATIONSHIP BETWEEN ORGANIZATIONAL CULTURE AND THE DEPLOYMENT OF SYSTEMS DEVELOPMENT METHODOLOGIES 1 By: Juhani Iivari Department of Information Processing Science University of Oulu P.O. Box 3000 90014 Oulun yliopisto FINLAND juhani.iivari@oulu.fi Magda Huisman School of Computer, Statistical and Mathematical Sciences North-West University, Potchefstroom Campus Private Bag X6001 Potchefstroom 2531 SOUTH AFRICA rkwhmh@puknet.puk.ac.za with a hierarchical culture that is oriented toward security, order, and routinization. IT managers critical attitudes of the deployment of methodologies in organizations with a strong rational culture (focusing on productivity, efficiency, and goal achievement) is also worth noting. Based on its empirical findings, the paper proposes a theoretical model to explain the impact of organizational culture on the deployment of systems development methodologies. Keywords: Systems development, software engineering, systems development methodology, organizational culture, competing values model, information systems developers, information technology managers Introduction Abstract This exploratory study analyzes the relationship between organizational culture and the deployment of systems development methodologies. Organizational culture is interpreted in terms of the competing values model and deployment as perceptions of the support, use, and impact of systems development methodologies. The results show that the deployment of methodologies by IS developers is primarily associated 1 Ritu Agarwal was the accepting senior editor for this paper. Gert-Jan de Vreede was the associate editor. Roberto Evaristo and Nancy Russo served as reviewers. The third reviewer chose to remain anonymous. Modern societies are increasingly dependent on software and information systems. The recent CHAOS report 2 estimates that total spending on systems development in 2004 was $255 billion in the United States alone. Although the pessimistic views of a continued software crisis and the high failure rate of systems development are exaggerated (Glass 2000), systems development continues to be challenging. Problems regarding the cost, timeliness, and quality of software products still exist. There have been several attempts to tackle these problems. The Information Systems and Software Engineering communities have witnessed a continuous stream of new systems 2 The Standish Group, available at http://www.standishgroup.com/. MIS Quarterly Vol. 31 No. 1 pp. 35-58/March 2007 35

development approaches, methods, techniques, process models, and related tools, even though their practical usefulness of methods in particular has been questioned (e.g., Baskerville et al. 1992; Fitzgerald 1996). It is our contention that systems development methods are significant for research and practice. They are repositories of codified knowledge on how to develop information systems and software artefacts (Fitzgerald 1998). They attempt to answer the central question of our discipline: How do we best design IT artifacts and information systems to increase their compatibility, usefulness, and ease of use (Benbasat and Zmud 2003)? Systems development is weakly addressed in the top IS journals (Vessey et al. 2002). Until recently, there has not been much research into actual use of systems development methods (Wynekoop and Russo 1997). Most earlier studies are descriptive (e.g., Chatzoglou and Macaulay 1996; Hardy et al. 1995) and do not attempt to explain the use and benefits of methods. It is only very recently that more explanatory studies have appeared (Hardgrave and Johnson 2003; Khalifa and Verner 2000; Riemenschneider et al. 2002). Although these studies analyze acceptance at the individual level, they clearly show that the deployment of systems development methods is a collective phenomenon. 3 Much of the existing criticism of systems development methods is also based on case studies (Kautz et al. 2004; Nandhakumar and Avison 1999; Wastell 1996) and therefore is not necessarily generalizable. Even though existing studies suggest quite consistently that many organizations claim not to use any methods (e.g., Chatzoglou and Macaulay 1996; Hardy et al. 1995), at least not rigorously or in their entirety (Bansler and Bødker 1993; Fitzgerald 1998; Kautz et al. 2004), our understanding of the contingencies under which methods are accepted is very limited. The purpose of this paper is to investigate the relationship between organizational culture and the deployment of systems development methodologies. Organizational culture may be one reason for the weak acceptance of methodologies. Although its significance as a source of organizational inertia is well known (Cameron and Freeman 1991; Schein 1985), the relationship between organizational culture and the deployment of systems development methodologies is unexplored territory (Leidner and Kayworth 2006). 3 In addition to perceived usefulness and compatibility, Riemenschneider et al. (2002) found subjective norm and voluntariness to be significant predictors of the intention to use a method. Hardgrave and Johnson (2003) report organizational usefulness, but not personal usefulness, as a significant predictor of the intention to use object-oriented methods. They also found subjective norm to be a significant predictor of organizational usefulness, but not of the intention to use. In view of the state of existing research, this paper is an exploratory, theory-building exercise. Methodologically, it is a quantitative survey. As we will argue later, there are no philosophical (Chalmers 1999) or methodological (Dubin 1978; Wallace 1983) reasons for preferring qualitative to quantitative research in theory building. Our question is, does organizational culture, when applied to IT departments, have any relationship to the deployment of systems development methodologies? Deployment here refers to perceived support for systems development provided by methodologies, the actual use of methodologies, and the perceived impact of their use on the quality of the system developed and the productivity of the development process. We will answer this question by developing a survey instrument based on the extant literature on organizational culture and SDM deployment. After reporting on the survey, we will analyze the relationship between organizational culture and SDM deployment. Based on insights from the empirical analysis, we will then proceed to building a theoretical model to explain the influence of organizational culture on SDM deployment. Organizational Culture, the Competing Values Framework, and the Deployment of Systems Development Methodologies Organizational Culture Organizational culture (OC) can be construed to cover almost everything in an organization: basic assumptions and beliefs, values, models of behavior, rituals, practices, symbols, heroes, artefacts, and technology (Gagliardi 1986; Hofstede et al. 1990; Schein 1985). Therefore it is understandable that it has several interpretations (Allaire and Firsirotu 1984; Czarniawska-Joerges 1992; Leidner and Kayworth 2006; Smircich 1983). Despite the differences, there seems to be an agreement that OC includes several levels with a varying degree of awareness on the part of the culture-bearers (Hofstede et al. 1990; Schein 1985). Schein, for example, suggests that the deepest level consists of patterns of basic assumptions that the organizational members take for granted without being aware of them. At the surface level there are artefacts such as the visible and audible patterns of the culture. The intermediate level covers values and beliefs, concerning what ought to be done. This paper focuses on this intermediate level of values, applying the competing values model (CVM) (Denison and Spreitzer 1991; Quinn and Kimberly 1984; Quinn and Rohrbaugh 1983) as a theoretical model of OC. 36 MIS Quarterly Vol. 31 No. 1/March 2007

Change Group culture Developmental culture Internal focus Hierarchical culture Rational culture External focus Stability Figure 1. The Competing Values Framework for Organizational Culture The major reasons for the selection of the CVM is that, as a quantitative model of OC, it is compatible with the survey research method selected for this study, it is well reported in the literature, and it has fairly short, validated measurement instruments for OC (e.g., Denison and Spreitzer 1991). Furthermore, to our knowledge there are not many alternative quantitative models of OC, the Organizational Culture Inventory (Cooke and Rousseau 1988) and the model of Hofstede et al. (1990) being notable alternatives. These two alternatives were far too complex, however, for the purposes of the present paper, both including more than 100 items required to measure culture. The CVM focuses on values as core constituents of OC. It is based on two distinctions: change versus stability and internal focus versus external focus (Figure 1). Change emphasizes flexibility and spontaneity, whereas stability focuses on control, continuity, and order. Internal focus underlines integration and maintenance of the socio-technical system, whereas external focus emphasizes competition and interaction with the organizational environment (Denison and Spreitzer 1991). The opposite ends of these dimensions impose competing and conflicting demands on the organization. Based on the two dimensions, one can distinguish four types of culture. The group culture (change and internal focus) is primarily concerned with human relations and flexibility. Belonging, trust, and participation are its core values. Effectiveness criteria include the development of human potential and member commitment. The developmental culture (change and external focus) is future-oriented, considering what might be. The effectiveness criteria emphasize growth, resource acquisition, creativity, and adaptation to the external environment. The rational culture (stability and external focus) is achievement-oriented, focusing on productivity, efficiency, and goal achievement. The hierarchical culture (stability and internal focus) is oriented toward security, order, and routinization. It emphasizes control, stability and efficiency through the following of regulations. Each of the cultural types has its polar opposite (Denison and Spreitzer 1991). A group OC, which emphasizes flexibility and internal focus, is contrasted with a rational OC, the latter stressing control and external focus. A developmental OC, which is characterized by flexibility and external focus, is opposed by a hierarchical OC, which emphasizes control and internal focus. The four are ideal types in the sense that an organization is unlikely to reflect only one type (Denison and Spreitzer 1991). CVM stresses a reasonable balance between the opposite orientations, although some cultural types may be more dominant than others. This imposes paradoxical requirements for effective organizations (Cameron 1986). Large organizations tend to develop a number of subcultures (Gregory 1983; Smircich 1983) instead of a single homogeneous culture. Recognizing this plurality, this paper applies CVM to IT departments, since they can be expected to be most closely associated with the behavior of IS developers and the deployment of SDMs. Deployment of Systems Development Methodologies The usage of systems development methodologies (SDMs) is a versatile concept. One reason is the ambiguity related to the MIS Quarterly Vol. 31 No. 1/March 2007 37

Table 1. Deployment of Systems Development Methodologies Aspect Dimension Methodology support 1. Perceived SDM support as production technology 2. Perceived SDM support as control technology 3. Perceived methodology support as cognitive & cooperation technology Methodology use 4. Maximum intensity of SDM use (vertical use) 5. SDM use across the organization (horizontal use) Methodology impact 6. Perceived impact on the quality of developed systems 7. Perceived impact on the productivity and quality of the development process term methodology. This paper uses the term to cover the totality of systems development approaches (such as the structured approach, information modeling approach, objectoriented approach, socio-technical design approach, etc.), process models (such as the linear life-cycle, prototyping, evolutionary development, and spiral models), specific methods (e.g., Yourdon s structured analysis, IE, NIAM, OMT, UML, ETHICS) and specific techniques in an organization. There are two reasons for this broad interpretation of an SDM. First, we wish to point out that the question is not only about the specific methods and techniques, but about more general approaches and process models. We contend systems developers may apply a methodology by following the goals, fundamental concepts, guiding principles, and principles of the systems development process of a specific systems development approach (Iivari et al. 1998) such as object orientation without strictly adhering to any specific methods. The second reason is related to the second source of ambiguity, the difficulty of defining and measuring SDM usage. Referring to Iivari and Maansaari (1998), one can distinguish explicit and implicit SDM use. Explicit use refers to consulting the method (documentation), while implicit use refers to the use of method knowledge after it has been learned and internalized, possibly years later. In an extreme case, implicit use may be an unconscious process in which method knowledge is intertwined with practical experience. Our broad interpretation of an SDM attempts to capture not only explicit SDM use, but also implicit use. Because of the difficulty of defining and measuring SDM usage, we will focus more broadly on SDM deployment, comprising methodology support, methodology use, and methodology impact (Table 1). The dimensions of methodology support are adapted from Henderson and Cooprider (1990), who identify three functional dimensions of IS planning and design aids such as SDMs: production technology, coordination technology, and organizational technology. They define the functionality of production technology as having a direct impact on the capacity of individual(s) to generate planning or design decisions and subsequent artifacts or products (p. 232). Coordination technology comprises control functionality and cooperative functionality. Control functionality enables the user to plan for and enforce rules, policies or priorities that will govern or restrict the activities of team members during the planning and design process (p. 236), while cooperative functionality enables the user to exchange information with another individual(s) for the purpose of influencing (affecting) the concept, process and product of the planning/design team (p. 236). 4 The two dimensions of methodology use are from McChesney and Glass (1993). Vertical use in an organization describes the intensity of method usage, while horizontal use describes the percentage of IS developers and projects using the methodology knowledge. The final two dimensions are adopted from Iivari (1996). Organizational Culture and Systems Development Methodologies Organizational culture forms the context in which systems development takes place. We formulated our question in the Introduction as follows: 4 We omit here organizational technology, consisting of two additional functionalities: support functionality, to help an individual user understand and use a planning and design aid effectively, and infrastructure, defined as standards that enable portability of skills, knowledge, procedures, or methods across planning or design processes. The support functionality can be interpreted as a meta-functionality in the sense that it supports the utilization of all the basic functionalities. One of the findings of the Henderson and Cooprider (1990) study was that support functionality was difficult for respondents to differentiate clearly (p. 244). The infrastructure component resulted from feedback during the study and its differentiation was not tested empirically. We see infrastructure functionalities such as standards as supporting cooperation. 38 MIS Quarterly Vol. 31 No. 1/March 2007

RQ: Does organizational culture, when applied to IT departments, have any relationship with the deployment of systems development methodologies? Despite the dearth of previous research into the problem, there are good a priori reasons to believe in a relationship between OC and the deployment of SDMs. Applying Schein (1985), Kekäle (1998) interprets OC as unconscious collective beliefs and assumptions that steer the values and through them the artefacts and actions of the organization, including the collective reactions as to whether a new approach or artefact is good or bad. This implies a conjecture that OC influences the collective reactions as to whether SDMs are considered good or bad, and consequently their deployment. The significance of OC as a source of organizational inertia is also well known (Cameron and Freeman 1991; Schein 1985), and there has been some interest in its influence on the acceptance of IT adoption, diffusion, and use (Leidner and Kayworth 2006). Applying CVM specifically, Cooper (1994) proposed that different information systems support alternative values, and that when an IS conflicts with the values of OC, implementation of the system will be resisted. Expanding CVM to comprise ethical culture, Ruppel and Harrington (2001) found that intranet implementation is facilitated by a culture that emphasizes trust and concern for other people (ethical culture), flexibility and innovation (developmental culture), policies, procedures, and information management (hierarchical culture). We are not prepared to put forward any a priori hypotheses about the relationship between organizational culture and SDM deployment for three reasons. The first reason is the richness of the concept of organizational culture, comprising symbols, heroes, rituals, values, and practices. Therefore, SDMs and their use can be conceived to be part of OC, as rituals (Robey and Markus 1984) that serve as a social defense against the anxieties and uncertainties of systems development rather than as an efficient and effective means of developing systems (Wastell 1996). Since OC and SDMs are not necessarily conceptually distinct, the suggestion that OC influences SDM deployment becomes problematic. To avoid this overlap, the paper selected CVM, which focuses only on values in an organizational culture. The second reason is that one can also conjecture that SDMs include certain cultural assumptions, and when these assumptions are incongruent with OC in an organization, SDM deployment is impeded. To our knowledge there is no previous research in this area, but Ngwenyama and Nielsen (2003) applied CVM to the analysis of the cultural assumptions of the capability in maturity model (CMM) literature and concluded that the design ideal of CMM reflects the rational culture, but becomes more hierarchical at higher levels of maturity. In view of the close affinity between CMM and SDMs, there are good reasons to believe that the latter may include certain cultural assumptions as well. The third reason is that CVM suggests that the effectiveness of an organization imposes paradoxical requirements in order to balance opposite cultural orientations. This implies that the assumed relationship between OC and SDM deployment may be either reinforcing or complementary. The former means that an SDM reinforces the existing OC and the latter that it complements it in some way. To exemplify the former, organizations with a hierarchical OC may use SDMs as a means of imposing security, order, and routinization. On the other hand, one can conceive that organizations with a developmental OC, for example, may also perceive SDMs as a means of imposing the necessary security, order, and routinization. Overall, this paper takes the view that the relationship between OC and SDM deployment is interactive and mutually constitutive, but the details of this relationship are still open. The purpose is to analyze this relationship using an exploratory research approach. Based on this exploratory analysis, the paper proposes a theoretical model with associated propositions and hypotheses that allow us to explain the findings. Research Design When analyzing the relationship between the OCs of IT departments and SDM deployment, we focus on the cultural perceptions of one occupational community (Van Maanen and Barley 1984), IS developers. The reason for focusing on the cultural perceptions of IS developers rather than of IT managers is to avoid associating culture with the IT managers view of the desirable culture to be imposed on the IT department. IT managers views of OC may represent an organizational ideology that they exercise in their normative control over IS developers (Kunda 1992). This ideology may differ radically from the OC perceived by IS developers. In the case of SDM deployment, we decided to study both IS developers and IT managers perceptions. One reason for this is the possible common method bias brought by a research design in which the same respondents (i.e., IS developers) assess both OC and SDM deployment. Our research design allows intergroup analysis in which OC is assessed by IS developers and deployment by IT managers. One should note, however, that the purpose of this study is MIS Quarterly Vol. 31 No. 1/March 2007 39

Table 2. Response Rate of Survey Number Distributed Number Returned Response Rate (%) Organizations 213 80 37.6 IS developers 893 234 26.2 IT managers 213 73 34.3 Table 3. Profiles of Responding Organizations Based on the IT Manager Data (N = 73) n % n % Business area Organization size Administrative services 3 4.1 1-50 employees 5 6.8 Finance/Banking/Insurance 11 15.1 51-200 employees 8 11.0 Software house/software consulting 5 6.8 More than 200 employees 60 82.2 Manufacturing 24 32.9 IT Department Size Retail/Wholesale 7 9.6 1 5 employees 17 23.3 Education 21 28.8 6 20 employees 23 31.5 Other 2 2.7 20 50 employees 12 16.4 More than 50 employees 21 28.8 not a systematic comparison of IT managers and IS developers perceptions (some of these have been reported in Huisman and Iivari 2006). The Survey This study is part of a larger survey of systems SDM use in South Africa, conducted in 1999. The 1999 IT Users Handbook 5 was used and the 443 listed organizations were contacted via telephone to determine if they were willing to participate in the study. In all, 213 organizations agreed to take part. A package of questionnaires was sent to a contact person in each organization, who distributed it. This package consisted of one questionnaire to be answered by the IT manager and a number of questionnaires to be answered by individual IS developers in the organization. The number of IS developer questionnaires was determined for each organization during the telephone contacts. The response rate is given in Table 2. Completed IT manager questionnaires were received from 73 organizations and completed IS developer questionnaires from 234 developers from 71 organizations. The total number of organizations was 80 and the number of responses from organizations with both IS developer and IT manager responses was 64. The distribution of IS developer responses per organization was skewed, so that only one developer questionnaire was received from 30.9 percent of the organizations, whereas 25.0 percent of the organizations returned five developer questionnaires. The maximum number of questionnaires returned by one organization was 11. The profiles of the participating organizations and individual IS developers are summarized in Tables 3 and 4. Due to problems with the mailing service in South Africa, it was not possible to analyze nonresponse bias based on a comparison of the latest replies, because one cannot be sure that the replies that arrived latest were written latest. 6 However, when we compared the sectoral composition of businesses in South Africa (ABSA Group 2000) at the time of the survey with the sectoral composition of business areas in our sample (Table 3) using the z-test for differences between two proportions, no significant differences were found. 5 The most comprehensive reference guide to the IT industry in South Africa, The 1999 IT Users Handbook is published by Computing S.A., TML Trade Publishing, PO Box 182, Pinegowrie 2123, South Africa. 6 Unfortunately, we were unable to obtain the envelopes with the date of mailing. 40 MIS Quarterly Vol. 31 No. 1/March 2007

Table 4. Profiles of Responding IS Developers (N = 234) n % n % Education Systems analyst 24 10.3 Senior certificate (high school) 39 16.7 Analyst/Programmer 93 39.7 Certificate or diploma 79 33.8 Programmer 36 15.4 University or technical degree 75 32.1 Other 29 12.4 Honors or Master s degree 36 15.4 Missing 3 1.3 Ph.D. degree 0 0.0 Experience in Systems Development Other 3 1.3 None 4 1.7 Missing 2 0.9 Less than 1 year 14 6.0 Title 1 2 years 21 9.0 IT manager 11 4.7 3 5 years 51 21.8 Project manager 25 10.7 5 10 years 53 22.6 Team leader 13 5.6 More than 10 years 89 38.0 Measurement The appropriateness and validity of the questionnaires were tested in two stages. First, six lecturers from the Computer Science and Information Systems Department at the Potchefstroom University for CHE tested the questionnaires. After some changes, they were pilot tested in practice at the IT department of an organization in Gauteng. The relevant part of the questionnaire is presented in Appendix A. All of the questions except those on OC and horizontal SDM use were addressed to both IS developers and IT managers. For the reasons explained above, only the IS developers were asked about OC and only the IT managers about horizontal SDM use. All of the measurement instruments, except that for OC, were specifically developed for the present study. Organizational culture was measured using the instrument suggested by Yeung et al. (1991). At the individual level, the scores for each of the culture orientations were computed as averages of the items included in the measure, and the cultural orientation at the organizational level was obtained as the average of the individual scores. Accordingly, even though the IT managers were not questioned about OC, the organizational culture of the IT departments they were heading was measured using the IS developer data. The details of the analysis of the measurement instruments are reported in Appendix B. The reliabilities of measures were tested using Cronbach s alpha, leading to the removal of a few items from the measurement instruments. Factor analyses of the measures for the seven dimensions of SDM deployment led to eleven factors. Factor analysis of perceived SDM support as production technology identified three factors: support for organizational alignment, support for technical design, and support for verification and validation. Factor analysis of perceived SDM support as cognitive and cooperation technology gave two factors: support for the common conception of systems development practice and support for the evaluation of systems development practice. Factor analysis of perceived SDM impact on the quality and productivity of the development process led to two factors: productivity effects and morale and quality effects, goal achievement, and reputation. Data Analysis The data analysis was performed using Statistica (version 5) software. Indices for the four organizational culture types for each organization were calculated as averages of the developers perceptions regarding the culture of that organization. For all other variables, individual developer and manager responses were aggregated separately to the organizational level by calculating the aggregated responses as means of the individual responses. Notes on Exploratory Surveys as a Research Method for Theory Building Because of the lack of previous research in this area, the phenomenon of the relationship between OC and SDM de- MIS Quarterly Vol. 31 No. 1/March 2007 41

ployment is poorly understood. There is no a priori theory to explain the phenomenon. 7 Consequently, this paper is of the theory-creating rather than theory-testing. Even though theory-creating research is sometimes associated with qualitative and interpretive research methods rather than with quantitative ones (Järvinen 2001), we do not see any philosophical (Chalmers 1999) or methodological (Dubin 1978; Wallace 1983) reasons why this should be so. Instead, we see the relationship between the purpose of a piece of research (exploratory/theory-creating versus confirmatory/theorytesting) and its methods as orthogonal. Qualitative research methods can also be used to test theories, as pointed out by Lee (1989), and quantitative research methods can be used to inspire theory building. Appendix C includes our argumentation of this point. This paper employs a survey as its research method in an attempt to create new theory. Based on survey data on the deployment of SDMs in South Africa, the paper proposes a theoretical model that allows interpretation of the empirical findings. The present study can be interpreted as an exploratory survey (Malhotra and Grover 1998). Even though Pinsonneault and Kraemer (1993) evaluate, based on an analysis of 122 surveys published by IS scholars between 1980 and 1990, that exploratory and descriptive surveys have been of moderate or poor quality, we do not interpret them as claiming that this is necessarily so because of inherent weaknesses of exploratory and descriptive surveys. Rather the question is about poor research design, sampling procedures, and data collection. Malhotra and Grover (1998) suggest 17 criteria for an ideal survey. The present paper clearly violates two of these (Criterion 10: Are pilot data used for purifying measures or are existing, validated measures adapted? Criterion 11: Are confirmatory methods used?). Two criteria (Criterion 6: Is content validity assessed? Criterion 9: Is construct validity assessed?) are only partially addressed. 8 In principle, these shortcomings in measurement could dilute the empirical findings of the present study. However, our results are not based on single measurements, but more holistic patterns and puzzles discernable in the empirical findings. Therefore, we see them as indicative enough to justify the theory-building exercise that will be reported later. Results To test the effect of individual culture orientations, regression analysis was used, considering each of the seven dimensions (with eleven factors) of SDM deployment as the dependent variable and the four indicators of organizational culture as the independent variables. The details of the regression analyses are reported in Appendix D and the results are summarized in Table 5, which lists the significant and almost significant regression coefficients identified (+ for positive and for negative, p # 0.05, and (+) and (-) for almost significant, p # 0.10). 9 One striking finding in Table 5 is the positive relationship between the hierarchical culture orientation and SDM deployment in the case of IS developers: the more hierarchical a culture is perceived to be, the more support SDMs are perceived to provide and the more they are used. The developmental culture is also found to have a positive association with SDM deployment, but not systematically so. Of particular interest, the more rational the cultural orientation, the more critical IT managers seem to be with regard to SDM support and impact. This is intriguing, since Huisman and Iivari (2006) found IT managers to have more positive perceptions of SDM deployment than IS developers. Regression analysis using a more comprehensive model, including the four cultural orientations, business area, IT department size, maturity of the IT department, innovation characteristics (relative advantage, complexity, compatibility, voluntariness), and percentage of time spent on new development, gave quite consistent results. The rational cultural orientation was negatively associated with deployment and the hierarchical cultural orientation positively associated with it (Huisman 2000). 10 The developmental cultural orientation did not exhibit any significant association with SDM deployment in this more extensive analysis. If we look at Table 5 with reference to the three aspects of SDM deployment, we can observe a decrease in the significance of the relationship between OC and SDM deployment in the case of IS developers from left to right. When we focus on the hierarchical cultural orientation only, the positive relationship is the most consistent in the case of perceived support 7 By theory, we mean an ordered set of assertions about a generic behavior or structure assumed to hold throughout a significantly broad range of specific instances (Weick 1989). 8 In the case of Criterion 6, only organizational culture is measured using an existing instrument. The other measures for the present study were developed based on the literature. Referring to Criterion 9, construct validity is assessed using only factor analysis for each construct separately. 9 The number of + and signs shows how many times a significant beta coefficient is found when the overall measures of the dependent variables are not included (e.g., in the last column of Table 5). 10 Note, however, that the sample size in this more extensive regression analysis was quite low relative to the number of predictor variables, only about five times the number of independent variables. 42 MIS Quarterly Vol. 31 No. 1/March 2007

Table 5. Summary of the Results of the Regression Analyses Group culture orientation Developmental culture orientation Hierarchical culture orientation Support as production technology (3) De: +(+) De: + Ma: +(+) Support as control technology (1) Support as cognitive & coordination technology (2) Vertical and horizontal SDM use (2) De: + De: (+) De: (+) Impact on the quality of the systems developed (1) Impact on the quality and productivity of the systems development process (2) Ma: (+) Rational culture orientation Ma: (-) Ma: (-) Ma: (-) Ma: (-)(-) De = IS developers; Ma = IT managers for systems development (two significant coefficients and one almost significant out of the six possibilities). There is one almost significant relationship in the case of SDM use (out of two possibilities) and none in the case SDM impact (out of three possibilities). This order, that SDM support perceptions are most affected by OC, then SDM use, and SDM impact least, is quite natural if one conceives an order of causality in which perceived SDM support influences SDM use and SDM use affects SDM impacts. On the other hand, the IT managers perceptions in the case of the rational cultural orientation behaved in just the opposite manner. IT managers are most consistently critical of the SDM impact (two almost significant relationships out of three possibilities), especially regarding the quality and productivity of the systems development process. There is one almost significant relationship in the case of SDM use (out of two possibilities) and two almost significant relationships in the case of SDM support for systems development (out of six possibilities). Discussion Theoretical Implications How can we explain the findings presented above? Let us start with the observation that the more hierarchical a culture was, the higher SDM deployment was reported by IS developers. One possibility is that SDM use is mandatory in the organization, and that its IS developers take this mandate more seriously than those in organizations with a less hierarchical culture. A second option is that SDMs as norm systems (Lyytinen 1986) are part of the social norms of the organization and the hierarchical culture affects the degree to which these norms are followed. Even though these two possibilities may explain the relationship between the hierarchical cultural orientation and SDM deployment, they do not easily explain why the same positive relationship was not discovered in the case of IT managers. It is also difficult to see how the mandatory nature of SDMs and social norms could explain IT managers critical attitudes in organizations with a strong rational culture. In the IT managers case, the expected outcomes related to SDM deployment provide a more natural explanation. The above considerations lead us to suggest the model in Figure 2 to explain our findings. 11 The model of Figure 2 makes a distinction between propositions and hypotheses based on their generality. Propositions are more general, whereas hypotheses are more bounded in time and space. We claim that the boundaries of the domain (Bacharach 1989; Dubin 1978) are very essential, especially when the conjectures concern human artefacts such as SDMs. Even though influenced by Dubin (1978) and Bacharach (1989), our use of the terms proposition and hypothesis differs 11 The dotted arrow in Figure 2 describes the feedback from SDM deployment. It is beyond the scope of the present paper to discuss it in detail. MIS Quarterly Vol. 31 No. 1/March 2007 43

Mandatoriness of SDM use Social norms concerning P2 SDM use P1, H1 P3, H3 SDM Deployment Perceived SDM support for systems development P4, H4 Organizational culture orientations P5 P6 H6a, H6b Relative emphasis placed on alternative values by an actor group Beliefs in SDM support for alternative values by an actor group P7 SDM use Perceived SDM impact Figure 2. The Theoretical Model from theirs. 12 For them, propositions exist between theoretical constructs and hypotheses between operational variables. Let us discuss the propositions and hypotheses of Figure 2 in more detail (Table 6). SDM use in an organization may be more or less mandatory. The word mandatoriness in Figure 2 is used consciously in contrast to voluntariness, in an attempt to capture the extent to which the desired behavior (SDM use in the present case) is made mandatory in the organization, whereas voluntariness (Moore and Benbasat 1991) is a more subjective view of the extent to which SDM use is perceived as voluntary. The hierarchical culture assumes that an individual will comply with organizational mandates (Quinn and Kimberly 1984). Therefore, Hypotheses H1 suggests that the strength of the hierarchical culture affects the extent to which mandatory SDMs are used. It is beyond the scope of the present paper to analyze in details how mandatoriness may affect 12 Based on Dubin (1978), it would be more appropriate to talk about laws of interaction and propositions in Figure 2. We are hesitant, however, to talk about laws in the context of behavioral sciences. SDM deployment. One could conjecture, however, that it affects the perceived voluntariness, which affects the relationship. Hypothesis H1 is in line with the significance of centralization and formalization in the implementation of innovations (Zaltman et al. 1973). Understandably, there is no prior research on the relationship between mandatoriness and SDM deployment. There are only a few studies on the impact of voluntariness on the acceptance of SDMs and related software process innovations (Green et al. 2004; Iivari 1996; Huisman 2000; Riemenschneider et al. 2002), and all of these report a significant negative relationship. Hypothesis H1 claims that this is especially so in organizations with a strong hierarchical culture. 13 13 According to Huisman s (2000) data at the organizational level, the correlation between the organizational average of voluntariness and horizontal methodology use was -0.32 (p # 0.05) in organizations with the hierarchical culture orientation # 3 (n = 45) and -0.65 (p # 0.01) in organizations with the hierarchical culture orientation > 3 (n = 15). The corresponding correlations in the case of vertical methodology use were -0.38 (p # 0.01) and -0.45 (p # 0.10). At the individual level, the correlation between voluntariness and vertical methodology use was -0.14 (p # 0.10) in organizations with the low hierarchical culture orientation (n = 139) and -0.29 (p # 0.05) in organizations with the high hierarchical culture orientation (n = 39). 44 MIS Quarterly Vol. 31 No. 1/March 2007

Table 6. Propositions and Hypotheses Propositions P1 Organizational culture orientations affect the extent to which mandatoriness of SDM use influences actual SDM use. P2 Organizational culture orientations affect the extent to which SDMs are made mandatory. P3 Organizational culture orientations affect the extent to which social norms concerning SDM use influence actual SDM use. P4 Organizational culture orientations affect social norms related to SDM use. P5 Organizational culture orientations affect the relative emphasis put on alternative values. P6 Organizational culture orientations affect the beliefs in SDM support for alternative values. P7 The relative emphasis placed on alternative values by an actor group and its beliefs in SDM support for alternative values have an interactive relationship with the deployment of an SDM. H1 H3 H4 Hypotheses The hierarchical cultural orientation affects positively the extent to which mandatoriness of SDM use influences actual SDM use. The hierarchical cultural orientation affects positively the extent to which social norms concerning SDM use influence actual SDM use. The hierarchical cultural orientation increases the number of social norms related to SDM use. H6a The rational cultural orientation has a negative impact on IT managers beliefs in traditional SDM support for productivity, efficiency and goal achievement. H6b The hierarchical cultural orientation has a positive impact on IS developers beliefs in traditional SDM support for control, stability and efficiency through following regulations. We do not propose any specific hypotheses corresponding to Proposition P2, but it can be conjectured that the desired behavior more easily becomes mandatory in organizations with a strong hierarchical culture and less so in organizations with a strong development culture, for example. Buenger et al. (1996) provide partial evidence for Proposition P2, reporting that the hierarchical cultural orientation (internal process value) was associated with vertical coordination. On the other hand, less mandatory action may be effective in a strong hierarchical culture, because the desired behavior will be implemented better (Hypothesis H1). Dysfunctions in strong mandatory action may also differ depending on the organizational culture. 14 Referring to Proposition P3 and Hypothesis H3, the hierarchical culture emphasizes control, stability, and efficiency 14 In Huisman s (2000) data, the mean for voluntariness in organizations with low hierarchical culture orientation was 3.33 and in organizations with high hierarchical culture orientation it was 2.57. The difference is statistically significant (p # 0.01). through following regulations and is oriented toward security, order, and routinization (Denison and Spreitzer 1991). SDMs are regulative norm systems (Lyytinen 1986). It is therefore likely that that following regulations, including SDMs, will be a natural form of behavior in organizations with a strong hierarchical culture and can take place without paying conscious attention to the underlying values of that culture (see Proposition P5 below). Hypothesis H3 is in line with the significance of formalization in the implementation of innovations (Zaltman et al. 1973). It also covers the significance of subjective norms (Fishbein and Ajzen 1975) as determinants of SDM deployment. H3 is more general, however, also covering situations in which following regulations or norms (such as the standard SDMs in the organization) is so natural that the members of the culture do not perceive that this is a question of social pressure to perform or not perform the particular type of behavior. Riemenschneider et al. (2002) and Hardgrave and Johnson (2003) have examined the significance of subjective norms as a predictor of the intention to use a method, the former reporting a subjective norm to be a significant predictor of MIS Quarterly Vol. 31 No. 1/March 2007 45

such an intention and the latter a significant predictor of organizational usefulness, but not of an intention to use the method. One explanation for this inconsistency in their findings could be that the two studies did not take the organizational culture into account. Hypothesis H3 suggests that a hierarchical culture may accentuate the significance of subjective norms as determinants of behavior. In the context of Proposition P4, our hypothesis is that organizations with a strong hierarchical culture develop more social norms related to SDM deployment, including SDMs themselves (Hypothesis H4). The close positive correlation between the hierarchical cultural orientation and formalization reported by Zammuto and Krakower (1991) supports the hypothesis. As argued above, we do not see that propositions P1 through P4 with their related hypotheses are effective in explaining the critical attitude of IT managers toward SDM deployment in organizations with a strong developmental culture. To explain this, we introduce values and actors beliefs into the SDM support for alternative values in Figure 2. According to Schein (1985), values and beliefs are central constituents of OC. The beliefs in Figure 2 are more specific, however, focusing on SDM support for alternative values. The model resembles the way in which attitudes are defined in the theory of reasoned action (Fishbein and Ajzen 1975), in that the relative emphasis on alternative values corresponds to an individual s evaluation of the consequences of his/her behavior and beliefs in SDM support for such alternative values. Differing from TRA, our emphasis on alternative values and beliefs in Figure 2 is more social, both being influenced by culture (Proposition P5 and Proposition P6). According to CVM, the four culture orientations (hierarchical, rational, developmental, and group culture orientations) influence the relative emphasis placed on alternative values by different actor groups (for example, emphasis placed on productivity and efficiency by IT managers versus IS developers). Proposition P5 allows for the fact that not all actor groups (e.g., IT managers and IS developers in our case) necessarily emphasize the alternative values equally, even though they may share the same organizational subculture. Despite the difference in the absolute emphasis on different values, P5 assumes that the direction of the influence of the culture will be consistent between the groups: the stronger the cultural orientation, the stronger the emphasis on the values of that orientation in each actor group. At the same time, the cultural orientations may also have an impact on the actor groups beliefs in the SDM support for alternative values (Proposition P6). A striking empirical finding in the present study is the negative association between the rational cultural orientation and SDM deployment in the case of IT managers perceptions, but not in the case of IS developers perceptions (see Table 5). Based on this finding, we propose Hypothesis H6a (Table 6). Figure 2 suggests two potential explanations for the difference between IT managers and IS developers in the case of a rational culture. First, IS developers do not emphasize productivity, efficiency, and goal achievement as values to the same extent as do IT managers. Second, IS developers do not view the current SDM support for these values as negatively as do IT managers. Both of these explanations seem plausible. It may also be that the strong emphasis on productivity and efficiency leads to a focus on short-run impacts, whereas the benefits of SDMs accrue more slowly (see Fichman and Kemerer 1993). In an extreme case, it may be a question of IT managers disappointment with SDMs when projects start to fall behind schedules. It is well known that SDMs are not very helpful in resolving these crisis situations, and that in these situations projects easily fall into a chaotic ad hoc style of systems development without any SDMs (Humphrey 1989). The critical attitude of IT managers toward SDMs in rationally oriented organizations seemingly contradicts the finding of Ngwenyama and Nielsen (2003) that the design ideal of CMM reflects a rational culture. It may be that SDMs differ from CMM in their cultural assumptions. On the other hand, it may also be that the underlying cultural assumptions of CMM, if evaluated by practitioners, would differ from the design ideals espoused in the CMM literature. A second striking empirical finding of the present paper is the positive relationship between the hierarchical cultural orientation and SDM deployment in the case of IS developers, but not in the case of IT managers (Table 5). Based on this empirical finding, we suggest Hypothesis H6b. Compared with Hypotheses H1, H3, and H4, this provides a complementary explanation (or possibly an alternative one) for the positive relation between the hierarchical cultural orientation and SDM deployment. One explanation for H6b is that SDMs are essentially norm systems (Lyytinen 1986), and following SDM regulations may be perceived as a means of supporting control, stability, and efficiency. As norm systems, SDMs may be perceived by IS developers to be part and parcel of the hierarchical culture. This is in line with Ngwenyama and Nielsen, who found that software process improvement models such as CMM reflect the hierarchical culture, especially at higher maturity levels. On the other hand, 92 percent of the responding companies in the present study were at the lowest maturity level (Huisman 2000). An explanation for this potential inconsistency may be that SDMs also reflect the hierarchical culture at the lower maturity levels when assessed by practitioners. 46 MIS Quarterly Vol. 31 No. 1/March 2007